elinks is hands-down the best console web browser (in comparison to
lynx, w3m and links) and under active development.
It is a fork of links with cleaner code, saner development roadmap (no
X11 support as in the recent bloated links versions) and a couple of
both advanced and handy features.
Among others, it has
Frame and table support; table support can be triggered on and off
with a hotkey when surfing a web site.
Can be scripted (locally, not remotely) with lua or guile; this
allows, for example, to enter & customize "smart addresses" like "gg
+ keyword" for google searches or
"wp" for wikipedia
browser tabs like in Mozilla, Konqueror etc.
typeahead jump to links like in Mozilla
Text area input fields of HTML forms can be edited with $EDITOR; for example, I am
typing this here in vim spawned from elinks (although this is a
little-known feature of classical lynx as well)
Optional html color(!) support (can be toggled, as all parameters)
Links can be numbered and accessed by number - as in lynx, but this
can be toggled during runtime
Support for SSL and form history
Javascript support (as a compile-time option) is on the way
High configurability/customizability, including all keyboard
shortcuts, through dialogues within the browser
And the best thing is: Despite all these features, the elinks executable
is still smaller than that of lynx (724k vs. 1.1 MB on my machine) and
links (902k).
In Germany and Austria, prices for new books are fixed, i.e. a new book always costs the same in every bookstore, including web bookstores like Amazon, so that bookstores don't compete on price. It does, however, not apply to (a) used books, (b) books where the rest of an edition is being discounted by the publishers, (c) books imported from other countries, (d) any other media, like music CDs, DVDs, software etc..
This law was made to protect small bookstores with chiefly literary, cultural and academic programs from the competition of bookstore chains (like Barnes & Noble in the U.S.) and mail order bookstores. While one might have different opinions about free markets and free pricing, this system indeed works as intended. Unlike in other Western countries, Germany and Austria benefit from a wealth of small quality bookstores in every town. In addition, there exists - since decades - a very efficient national book wholesaler system, so that any bookstore, regardless the size, can get any available book for a customer usually on the next day (if it's not on stock in the store already). Despite all this, Amazon still managed to establish a hugely profitable business in Germany for various reasons - the comfort of browsing an online catalogue and because they offered, for the first time in Germany, an efficient way of ordering English-language books.
The court decision simply maintains the fixed book price law for Ebay sales of new books by commercial traders. It does not apply to Ebay sales of second-hand books.
Several years ago, I wanted to start a grassroots campaign against MS
Word as an exchange document format as well and registered the domain
no-doc.org for that purpose. My idea was to gather wide-ranging support
not only from computer programmers, but also from non-technical people
like professional writers, and spread a "no-doc.org" logo (perhaps with
a crossed-out Microsoft Word file icon) as a popular image on stickers,
t-shirts, website banners etc..
However, I gave up on that project simply because, unlike for example in
the case of gif vs. png images, there is no easy
replacement to be advertised and offered to non-technical people:
RTF
is nothing but.doc in ASCII encoding, but otherwise it's also a format
defined and controlled by Microsoft (and whose newer versions I believe
are undocumented).
.swx, the OpenOffice/StarOffice format is currently not supported by
any other program.
HTML doesn't preserve important formatting information like
footnotes and is too inconsistently implemented across applications
DocBook and LaTeX are semantically structured formats and hence
not capable enough of supporting documents formatted with no structured
semantics; you can use them to create output in other formats, but you
can't automatically transform arbitrary formatting into them.
PDF is a write-only format
plain ASCII text offers not no formatting at all and is incompatible
across platform/language-specific special character and CR/LF encoding
In other words, a format that is open across applications and platforms,
sufficiently powerful in its encoding both of typographic (font settings
etc.) and structural (footnotes etc.) layout and widely supported by
mainstream word processors (and be it only everything but MS
Word) doesn't exist. As long as this doesn't change, for example with
OASIS' current efforts to standardize an open office document format or
large cross-application support for the OpenOffice file format, any
"no-doc" advocacy is elitist and doomed to alienate even people who
might be sympathetic for political reasons.
But if anyone wants to seriously do a grassroots campaign against using
Microsoft's proprietary file formats, I am happy to transfer the
no-doc.org domain to them for free.
Not if you use "enterprise" distributions, rely on support contracts,
need staff training, inhouse development for migrating data and
applications, or if you rely on proprietary software that runs on top of
the OS (the way the film industry uses GNU/Linux as graphics workstation
OS, for example).
But it's true that desktop GNU/Linux is good enough to have
commoditized the product "desktop operating system" as it is defined by
the standard system CD of MS Windows or MacOS X. Today, buying
Windows licenses only makes sense if one needs something for running
software on top of it that doesn't exist for GNU/Linux. Unlike in the
days of the early MacOS or Windows 3.x, the raw desktop for itself has no
proprietary product value anymore because KDE and Gnome have
commoditized it. With the exception of Web browsers, GUI
applications however are hardly commoditized yet. It may take a very
long time until this point is reached, or it even might not be reached
at all because of outside obstacles like software patents (on multimedia
codecs etc.) and DMCA-imposed interoperability restraints.
- Distros like Xandros "just work"
Can be said about Windows 2000 and XP as well.
- Linux is secure from worms, trojans, viruses
This is a dangerous delusion. GNU/Linux has none of that security
by design, but rather by the obscurity of not being a highly uniform
mainstream desktop OS. There have been enough security holes in critical
components of GNU/Linux (including the kernel itself) for malware to do
as much harm as under Windows, and I guess a skilled attacker can crack
almost any non-professionally administrated Linux box.
- Linux runs on modest hardware
Not if you run KDE, Gnome and Openoffice (as in a desktop
system). In fact, a Windows 2000-based desktop system is faster and
resource-friendlier in comparison.
- Linux is less complex and thus more stable
Nonsense. The combo of Linux kernel + OS daemons + GNU/shell userland
+ XFree86 + Gtk/Qt + KDE/Gnome middleware is just as complex as
Windows.
- Linux has a "cool" factor missing from Windows
No argument.
- The IT world's view of Microsoft as "evil" is percolating down to
the general public
No argument.
- Linux now comes with a sufficient set of applications for most
common purposes
No, if you speak of the desktop. You can deploy a Linux-based desktop
only in very well-defined (or limited) usage scenarios:
- Use of Internet clients (Mozilla): fully equivalent to Windows-based
solutions
- Office suite (OpenOffice vs. MS Office): only against tradeoffs in Microsoft
compatibility, speed, features and desktop integration
- Graphic design (Gimp vs. Photoshop): only against tradeoffs, suitable
only for screen design
All other free desktop applications which currently exist for GNU/Linux
are either small utilities (CD burning GUIs and the like), or they are
ambitious, but not yet complete and not available in production quality
(Scribus, Ardour etc.).
- Linux applications are more stable and simpler than Windows' ones
You don't mean the shell and the commandline userland, don't you? Which
applications do you mean if you speak of the desktop, and not of mutt, vim,
slrn etc.? (The GUI applications mentioned before are all cross-platform
*nix/Windows btw.)
What you write are the clichés Windows people have in mind when they
switch to GNU/Linux. Most of the rely on a false extrapolation from
GNU/Linux as a Unix-like commandline and server operating system (which,
if properly administrated, indeed is rock-solid and resource-efficient)
onto GNU/Linux as a desktop OS.
I doubt that. Apple already has its own presentation program and
therefore no use for KPresenter. Despite having greatly processed, KWord still doesn't feel like a
production app, has an overally awkward user
interface, much inferior import filters to OpenOffice and, at least in the
past, insufficient stability. It is therefore not nearly as attractive
as the base of
a potential "iWord" as KHTML was for Safari. Abiword, which would be an
alternative, also lacks reliability and compatibility, and OpenOffice
is a monster that Apple would avoid for the same reasons it avoided
Mozilla.
Also, KWord is built on top of KDE's
component/toolkit architecture that is a world apart from MacOS X
Carbon/Cocoa API. While Qt allows a native port to Aqua, it does
not offer a native port to Carbon or Coca, and Apple is unlikely
to establish a third desktop API on its platform just for the sake of
getting a functionally rather limited word processor that, at the
moment, has no dramatic advantage
over the old Claris/AppleWorks offering.
And keep in mind that for Safari, Apple just used the
engine (KHTML) of a free program, not the GUI application (konqueror)
itself, in the same way it put its own (proprietary) GUI on top of Mach
and BSD. From I experience, I doubt
that KWord and Abiword are, in their present state, as attractive as "engines" as BSD and KHTML
were. If it all, Abiword seems a more likely candidate since it's
designed as a cross-plattform application and, quite in opposition to
KWord, focuses on getting base
functions and usability right before acquiring more nifty/hackerish
features such as frame-based page layout and importing PDF files.
What makes your scenario very unlikely in the end are licensing issues.
KOffice and Abiword are GPLed code and thus would require Apple to release
any program based on them under the GPL. Which doesn't fit
to the company's successful tactics of
putting slick, but proprietary GUIs on top BSD- or LPGL-licensed hacker
code like BSD and KHTML.
A GPLed "iWord" that could be ported back to Linux and even Windows
would, unlike the current i-apps, be no exclusive selling point for MacOS X.
-F
Re: my-kingdom-for-a-wordperfect-import-filter
on
Koffice 1.3 Released
·
· Score: 2, Interesting
> I'm happy with my Linux system right now. It supports all my hardware > and gives me a nice desktop.
Then there's no reason for you to switch. If you would instead not be happy with your system and find that FreeBSD runs better on your hardware, than this would be a reason to go through the hassle of switching your OS.
After all, the differences between a GNU/Linux and a *BSD system are practically user-invisible on the level of the desktop interface. Both are, as a matter of fact, state-of-the-art Unix, but may not be called so because of trademarks and expensive certifications (which, contrary to popular make-belief, are not owned by SCO, but by the Open Group, formerly X/Open).
The differences mainly concern the kernel (partitioning schemes and filesystems, hardware drivers, module handling, packet filters, sound and multimedia subsystems). Userspace differences in the init system, package management and base OS/distribution tools are not bigger than those between two GNU/Linux distributions. Slackware or Gentoo users might even find Free/Net/OpenBSD more familiar than RedHat or SuSE. There are subtle, but sometimes crucial differences in the commandline userland between GNU/Linux and *BSD though, unless one installs the GNU file and text utilities on *BSD and uses them as default (which is easy and supported by the package management of all three free BSDs). The KDE/Gnome/XFCE desktops act in a completely transparent manner, with no visible differences, on top of GNU/Linux and *BSD.
Generally, the Linux kernel is best suited for a desktop system because of its more advanced sound (ALSA) and video (video4linux) subsystems, support for a wide range of desktop hardware gadgets (Webcams, graphic tablets and the like), filesystems (including everything from Amiga to Acorn...) and hardware-accelerated video card drivers (DRI/DRM, although I read similar features are on the roadmap of FreeBSD and NetBSD), and, since kernel 2.6, kernel preemption and low-latency functions. However, FreeBSD should come close to Linux as it is optimized for the x86-PC architecture and is a very good performer. NetBSD is, from my own experience of running it as a secondary OS, not as fast, but still surprisingly good for an OS that is developed with portability (and hence abstraction/clean interfaces vs. optimization) as its prime objective. IMHO, it is (very) roughly comparable with Linux kernel v2.0 in terms of performance and desktop computing friendliness. Installation of NetBSD is a bit difficult (more so than even Debian), and the necessity of creating classical BSD disklabels for every storage media to be mounted can be highly annoying on a desktop system (for example, if one wants to quickly mount someone else's USB stick). OpenBSD is, IMHO, a bad choice for a desktop system unless security and crypto features are the main requirements. It is not a good performer at all (and not being developed with performance as a main goal).
Kernel-wise, FreeBSD's chief advantage over Linux used to be better responsiveness under high system load and better virtual memory management (which both gave/give FreeBSD an edge over Linux on servers rather than on desktops). This advantage has gradually decreased through substantial low-level improvements in Linux 2.0, 2.4 and now 2.6, which AFAIK has lifted quite a bit from FreeBSD's advanced VM management. Maybe Linux 2.6 is now on par, but still I wouldn't be surprised if FreeBSD (and also NetBSD) would be more mature in this field. (For example, I never succeeded in bringing down my two NetBSD boxes with a fork bomb.)
File selector boxes are a legacy of the early MacOS until version 6.x,
which was single-tasking and didn't allow to switch between several
applications running parallel. In fact, a file selector box is nothing
but a miniature replica of a graphical file manager (like the MacOS
finder, the Windows Explorer, konqueror, nautilus, rox etc.). The more
"functional" file selectors got, the more bloated and redundant
vis-a-vis the file manager they became.
It would make more sense IMHO to abolish file selectors altogether and
instead throw users into their preferred file manager for opening files.
All it would need is a freedesktop.org standard protocol for file
manager/application interaction and perhaps a $FILEMANAGER environment
variable. (Theoretically, $FILEMANAGER could then also be a
shell in a terminal.)
Instead of re-inventing the wheel, people should just pick the
TEI (respectively TEI Lite) SGML/XML DTD of
the Text Encoding Initiatve.
For
those who haven't heard of it yet: TEI is an open SGML/XML format created for
electronic editions of literary texts. It is as comprehensive and
well-designed for text philology as DocBook is for technical
documentation. The only drawback is that it is, like DocBook,
very comprehensive and accurate in its markup tags (fulfilling
all needs of academic editions of historical texts), so that for
average readers, the trimmed-down TEI Lite DTD should do the job.
For e-literature collections created by professional philologists - such as the
Victorian
Women Writers Project, TEI already is the standard text format.
Thanks to the SGML/XML toolchain, TEI sourcecode can, like DocBook, of
course be painlessly transformed into HTML, txt, RTF, PDF etc. (TEI is,
btw., also being mentioned in Eric S. Raymond's quite useful DocBook
Demystification HOWTO.)
The first thing UserLinux needs to fix is its own name. Given that it
will not be end/home user distribution, but a business OS
designed to compete with the expensive "enterprise" offerings of RedHat
and SuSE, it should be better called "Free Enterprise Linux". (The
term "free enterprise" would also communicate to corporate people what
"free" in "free software" is about.)
If UserLinux was an end user-oriented distribution, it surely had to
pick KDE instead of Gnome, since KDE is the more integrated and stable
GUI and is less messy in the architecture underneath (while Gnome/GTK
has the lead in 3rd-party applications and, since recently, UI
polish).
But for a "Free Enterprise Linux", there must not be any hidden costs
for enterprise software development. This demands that libraries and
SDKs should, where possible, be LGPL- or BSD-licensed, and not GPLed
with for-pay-exceptions (like in Qt and MySQL).
Of course, the question remains if, due to its proprietary-friendly
licensing and relatively conservative (=stable) design process, FreeBSD
wouldn't be the better "Enterprise Linux" anyway. After all, the GPLed Linux
kernel could be ditched in favor of a BSD kernel with almost the same
arguments the UserLinux project now ditched the GPLed KDE libraries in favor of
the LPGLed Gnome libraries.
But since Linux is all
the hype even where it doesn't make too much sense (like in PDAs, for
which Minix would be much better suited), it's good that the "UserLinux"
project attempts to prevent that commercial distributors do the same
horrible mistakes with Linux and their "enterprise" distributions the
proprietary Unix vendors made in the 1990s.
-F
Misleading / questionable title
on
Linux Power Tools
·
· Score: 5, Interesting
The title "Linux Power Tools" falsely suggests that the book is a sequel or update to O'Reilly's "Unix Power Tools". Unlike the book reviewed here, "Unix Power Tools" is not about configuring KDE and such stuff, but a wizard's guide to & treasury of classical shell tools, arguably the best Unix user book ever written, the bible of the commandline, the ultimate celebration of design philosophy genius behind Unix. In that light, calling the reviewed book "Linux Power Tools" is pure blasphemy.
Have you been successful in running this drive under Linux? I am suspicious that it is identical to a BTC drive with exactly the same specs which, as my own experience and (unfortunately, two days too late) a test of the German "Linux User" magazine told me, does definitely not work under Linux.
...true, but DVD-Rs are more compatible to standalone DVD players and
older DVD-ROM drives. So far, I have been out of luck with all my DVD+Rs
on them, but successful with DVD-Rs. Dual standard DVD burners therefore
are not just a marketing gag, but of serious practical value.
Some of them not. I had bad luck with a BTC lowcost DVD +/-
that had very good specs (4x DVD-R/+R, 32x CDR burning), but worked
neither with dvdrecord, nor growisofs, nor the nonfree dvd-capable
variant of cdrecord. I wouldn't be surprised if the drive was identical to the LiteOn
DVD+/- that has identical specs.
Fortunately, I was able to exchange the BTC against a marginally more
expensive LG drive (4040B). While it burns CDRs only at 24x and CDRWs
at 12x, it worked out of the box and like a charm with all programs
under GNU/Linux.
A nice plus of the LG is that, besides DVD+-R[W] and
CDR[W], it also reads and writes DVD-RAM.
Unlike the other writable DVD
and CD formats, DVD-RAM is true random access optical removable storage
(similar to an MO cartridge, not just a userspace simulation of random
access like Mount Rainier packet writing). It works as standard IDE
device without any specific drivers or software frontends, so that
DVD-RAM media can be partitioned and formatted with ext2/xfs/reiserfs
under Linux just like a removable hard drive. Since it is possible,
for example, to rsync directories to DVD-RAM and since the
media are affordable (~$13 for a double sided 2x4,7 GB disk), it strikes me
as a near-ideal backup solution.
ratpoison, a
keyboard-controlled "anti-desktop" wm largely modelled after GNU screen,
has this feature in CVS since about one week.
However, it was added without patching the wm itself and thereby bloating
its code. Instead, since ratpoison can be fully controlled via the commandline,
the "expose" functionality (to be found in the "contrib" directory as
"rpshowall.sh") was written as an external shell script
which tells ratpoison to split frames in a certain way. Through
ratpoison's freely definable keybindings, the script can be used like a
built-in function/command of the wm.
Beg to differ, my own predictions:
on
Linux in 2004?
·
· Score: 5, Insightful
Which Linux application area do you believe will grow the fastest in 2004?
Conventional, but probably true answer: servers. There are still many
companies running standard services like mail, web etc. on proprietary
operating systems (Sun, Microsoft) in a time where it makes no
whatsoever sense anymore. With kernel 2.6, Linux will gain acceptance as
a high-end Unix replacement and be deployed wherever older server
installations need to be replaced.
Will 2004 *finally* be the year when Linux makes significant in-roads
on the desktop?
No. The desktop UI is still too inconsistent across KDE/Qt, Gnome/GTK,
Mozilla/XUL and Openoffice and still offers no viable alternative to the
commandline when it comes to system administration/configuration.
I predict that in 2004, attention will move away from
KDE and Gnome as all-in-one-solutions. Instead, it will be finally
accepted as reality among developers and users that different GUI
APIs will continue to coexist, and that efforts should be made to
standardize the protocols and user interfaces across the APIs.
For the future of GNU/Linux and *BSD on the desktop, freedesktop.org
will be much more important than kde.org and gnome.org, but it could
take five-ten years until the difference between a KDE/Qt, GTK/Gnome,
Tcl/Tk, Fltk program will be as irrelevant to users as the difference
between a Carbon and a Cocoa app on MacOS X or as that between a
Microsoft MFC program written in C++ or an OWL-based program written in
Borland Delphi for a Windows user.
Once this level of standardization is reached, the importance of
all-in-one desktops like KDE and Gnome could dramatically decrease,
since users instead could combine components like taskbars, window
managers, file managers and system menus at will. (Which, thanks to
freedesktop.org, is already possible: fspanel / fbpanel / suxpanel / the
xfce4 panel can be used as drop-in replacements for the Gnome panel, rox
/ xffm4 as drop-in replacements for Nautlius, and the list of
freedesktop.org-compliant window managers suitable as replacements of
metacity / kwin is endless.)
However, it will take yet another five years until 2013 or 2014 that a
standardized Unix/GNU/Linux/BSD desktop will allow developers of system
components (like sendmail/exim/Postfix/qmail, lpr/cups, Samba,
Grub/Lilo...) to write GUI configuration panels for their own software.
At the moment, desktop projects like KDE, Gnome/Ximian and Webmin can
only provide insufficient configuration wrappers around low-level system
tool; the only sane solution is that such GUI configuration panels are
provided by the original component developers in sync with their release
schedules, and will work consistently on any GUI configuration (as
opposed to the present situation where a configuration panel would have
to be provided in separate versions for KDE, Gnome, XFCE, webmin and
what-have-you). Only at this point, GNU/Linux will be able to replace
commercial end-user GUI operating systems on a large scale and be
accessible to home users.
Which distributions will show the greatest growth in 2004?
Contrary to what Eric S. Raymond says: The unclear situations
of RedHat/Fedora and SuSE (after it has been bought by Novell) could
create a strong push towards Debian as the standard binary
(GNU/)Linux distribution. The Debian core distribution could become a
de-facto-replacement of the disappointing "Linux Standards Base (LSB)",
as more and more (commercial and community) distributions will be based
Debian. Knoppix, Lindows and, in the near future, User Linux are
prime examples. Debian itself will gain more acceptance in the
mainstream and among new users as soon as it will ship with the new
installer.
Given the record of Netscape/Mozilla's, StarOffice/OpenOffice's
and Apple Darwin's transformation of corporate into public development
projects, I doubt that RedHat/Fedora will ever become a true community
project. It is also being overlooked that the equation RedHat=Linux
is specific to the U.S. only
...otherwise he wouldn't prefer MacOS X on a G5 to a free Unix clone like
Debian GNU/Linux or, to take a derivate of his own work, NetBSD on the
same hardware. While OS X provides the more powerful (yet proprietary)
GUI application layer, it is IMHO vastly inferior for being used the
classical Unix way:
The Mach+BSD server design is a kludge creating unneccessary bloat,
complexity and performance overhead without exploiting any of the
potential advantages of a microkernel design like better portability or
Hurd-style hack value like filesystems running as daemons in userspace
etc.
In any case, Linux 2.4 and all the more 2.6 should beat, in terms of
performance and scalability, the crap out
of MacOS X' combination of vintage Mach with vintage BSD and a bloated
GUI on top
Debian and NetBSD don't have compatibility bloat like the "Classic"
virtual machine, m68k-CPU-Emulation and "Carbon"-API in MacOS X
They have much cleaner filesystem layouts
than OS X with its inconsistency of Unix directories (/bin,/etc) which
are hidden on the GUI level and application folders inherited from
NextStep
They have a more consistent and robust configuration system than
MacOS X with its horrible Registry-like "Netinfo" database that
replaces some, but not all configuration files in/etc
They come with a more complete (and especially in Debian's case
thanks to GNU) powerful set of classical Unix commandline applications
For the software which is not installed by default, they have consistent package management while MacOS X has a
number of simultaneous/incompatible package managers and databases which
don't know each other's dependencies: MacOS X install images, fink,
GNU/Darwin, BSD-style pkgs/ports...
They install programs like vim, mutt, shells etc. with sensible default
configurations while I find the commandline userland and MacOS X almost
unusable they way it is configured out of the box
For someone who primarily works on the commandline and needs graphical
programs only for the occasional web browsing, graphics/pdf and video
viewing (for all of which excellent free, X11-based solutions like
mozilla-firebird and mplayer do exist), MacOS X offers no advantages
over a GNU/Linux or NetBSD system in which all the system- and
commandline-level things are done cleaner and better. So it seems Bill
Joy doesn't write in vi and work the Unix way anymore, otherwise he
would have better things to say about Linux.
This comment must have been written by someone (and modded up by someone else) who doesn't know much about XFCE. XFCE is not just a window manager, but a fully integrated, mouse-configurable desktop including wm, panel (with panel applets), taskbar, pager and graphical file manager (including Samba browsing support), extensive drag'n'drop capabilities including for printing, central configuration menus including sound and mouse setup. It's based on Gtk 2.x and freedesktop.org standards (and thus with a high degree especially of Gnome and KDE interoperability).
Think of XFCE as a desktop environment without the redundant middleware layers (DCOP/KParts/arts etc. in KDE, Corba/Bonobo/esd/Gconf etc.i in Gnome) that make both KDE and Gnome bloated and slow, and which are hardly used by third party applications outside main Gnome and KDE distributions at all.
So it amazes me that the previous commentator thinks XFCE is "not a desktop". On the contrary, XFCE is a desktop done architecturally right, similar to, for example, the desktops of AmigaOS, RiscOS, Macintosh Classic and BeOS. While both the XFCE panel (with its legacy to the user interface of the CDE panel) and new file manager could still need some usability improvement, the architectural foundation is excellent.
XFCE is also the proof that a X11- and GNU/Linux-/BSD-based desktop computer can be as fast and efficient as one would normally expect from a Unix-like system. In other words, it's as fast as a basic window manager setup with Window Maker/icewm/fvwm2 while providing a fully integrated desktop that doesn't require users to run the shell or edit configuration files.
(A prominent XFCE user and supporter is, btw., Alan Cox.)
For running an MTA on a desktop/client PC, I strongly recommend solutions like Nullmailer or, for computers with permanent Internet connectivity, ssmtp. Both work as just local gateways/bouncers to a remote SMTP server; they don't open any network ports and thus prevent remote exploits/attacks/spam relaying by design. Nullmailer offers local spooling (important for dialup connections) while ssmtp bounces everything immediately to the smarthost. Both are very small (ssmtp: 22k, nullmailer-send: 25k), ridiculously simple to configure even for people with low administration skills, both provide sendmail-compatibility to work with MUAs like mutt.
(Offtopic: A similarly nice, elegant solution for desktop/clients PC printing is pdq, which unlike lpd and cups runs only as a local spooler without opening a network port, and is lean (65k), dead-simple and functional. With nullmailer/ssmtp & pdq, I managed to close all ports (except of course SSH) on my two desktop PCs under Debian GNU/Linux without any firewalling. AFAIK, Debian is the only OS offering all the aforementioned pieces of software as part of its main distribution.)
Translation (German original contains a number of colloquial expressions which don't translate 1:1 into English): "What can I say? A top-notch piece of engineering. I am drooling!"
> The wording of the GPL is still valid. The GPL wouldn't "dissapear".
> Similarly, if you like the OPL, keep using it. It's still a perfectly valid, legal license.
Yes, but unmaintained legal code is as problematic as unmaintained program code. No organization will enforce these licenses or, if necessary, defend them in court; nobody will update them if new legal or technical conditions make it necessary (as in GPL v2.0 vs. GPL v1.0 vs. the upcoming GPL v3.0). Which renders opencontent.org's licenses worthless IMHO.
As a lecturer in the humanities and net activist who has been
evangelizing open content internationally in lectures, papers and as the
moderator of congress panels since 1999, I feel like being slapped into
my face. It is terrible if you educate people about open content and the
necessity of copylefting public information resources, pointing them
again and again to opencontent.org and their licenses and now see that
reference dissolve.
It is especially not funny to see the Open Publication License go away.
It had a considerable momentum among book publishers - being used, among
others, by O'Reilly and the Bruce Perens book series of Prentice Hall. I
myself put all my papers under the OPL, encouraged other people to do so
as well, and now feel severly f*cked and betrayed by this move. The
instability and unreliability now associated with open content copylefts
could severely damage the whole movement. As someone who managed to convince
a large German public library to release its online content under the
Open Content License, I am severely pissed & awaiting to take the
beating for opencontent.org's irresponsibility.
The Creative Commons licenses, in my view, are not an alternative
because they are too many and incompatible to each other, thus creating
confusion and preventing
exchange between work copylefted under its terms. What's still worse is
that most Creative Commons licenses are not free in the sense of the
Free Software definition of the FSF, the Debian Free Software Guidelines
or the Open Source Definition.
I urge the initiator of opencontent.org to keep the website alive, and
if only as a central link repository to other sites, and provide a
smooth/sensible upgrade path from the Open Content License and the Open
Publication License to particular Creative Common Licenses, for example
by developing a license which would simultaneously be "Open Publication
License v2.0" and "Creative Commons License foo". Given the amount of
work that already circulates under either the Open Content License or
the Open Publication License, anything else would be utterly
irresponsible.
Imagine the FSF suddenly abandoning/stalling the GPL in favor for someyet-unwritten different license, leaving ten thousands of Free Software
developers in the legal lurch & betraying their trust. What is an
unlikely horror scenario for free software is now the reality of open
content.
Bravo, opencontent.org, Microsoft, the RIAA, the MPA, SCO and all other old copyright
regimes now have another reason to cheer and point at copyleft culture
as immature, unreliable, not viable for serious publishing, etc..
Please wake up and release that you have taken up a
responsibility which you cannot so easily throw away!
To my knowledge, all Creative sound cards incl. Live and Audigy just
all other consumer sound cards on the market use an internal sampling frequency of 48 KHz (DAT standard) and have to resample their output for CD-standard 44.1 KHz output.
This is why audio professionals use much more expensive sound cards like the RME Hammerfall even when they only need analog stereo output.
Do new consumer cards solve this problem, or do they layer up useless extra features?
A major improvement in FreeBSD 5.x over 4.x is the new modular init. Instead of one monolithic script (classical BSD) or several scripts in a symlink farm with manual sorting and dependency resolution (SysV / Debian, RedHat, SuSE...), it uses an internal automatic sorting and dependency resolution comparable to apt-get or modprobe on GNU/Linux. I would like to see mainstream adoption of this in the GNU/Linux world of this. To date, Gentoo Linux is the only distribution offering and supporting this excellent feature.
Minix = great OS on old hardware/embedded
on
The End Of Minix?
·
· Score: 4, Informative
I fail to see why dropped XFree86 support would kill Minix - as it doesn't make much sense to run Minix on hardware powerful enough to run an X server.
After trying several alternatives (Small Linux with kernel 1.0.9, FreeDOS, V2OS), I found Minix an excellent OS for a laptop with a 386sx/20Mhz CPU with 2 MB RAM and 40 MB hard disk. On this hardware, Minix gives me a fast-running, stripped-down Unix shell environment comparable to Busybox (or respectively, Linux rescue floppies like Tomsrtbt), with a decent vi clone (elvis-tiny), a C compiler (!) and full man page documentation. It is a very sound choice for turning fleamarket hardware into a word processor with the Uni toolchain (grep, sed, sort, make etc. - for a full list see here). Exchanging data between Minix and other OSes is a snap thanks to Minix-fs support in the Linux kernel and the availability of mtools for Minix.
With its academic background, Minix is a very cleanly designed, BSDish OS. Its major drawbacks are lacking job control (suspending, backgrounding and foregrounding processes), an almost DOS-like limitation on the length of file names and, unless you use the MinixVM fork, no virtual memory.
Since Minix has been put under the BSD license since April 2000, I wonder why nobody has made an effort yet to port it to embedded systems (PalmOS PDAs with Dragonball CPUs, for example, should be an ideal target). Minix should be much better suited for many embedded applications than the much more complex Linux kernel.
Slashdot Mirror
It is a fork of links with cleaner code, saner development roadmap (no X11 support as in the recent bloated links versions) and a couple of both advanced and handy features.
Among others, it has
- Frame and table support; table support can be triggered on and off
with a hotkey when surfing a web site.
- Can be scripted (locally, not remotely) with lua or guile; this
allows, for example, to enter & customize "smart addresses" like "gg
+ keyword" for google searches or
"wp" for wikipedia
- browser tabs like in Mozilla, Konqueror etc.
- typeahead jump to links like in Mozilla
- Text area input fields of HTML forms can be edited with $EDITOR; for example, I am
typing this here in vim spawned from elinks (although this is a
little-known feature of classical lynx as well)
- Optional html color(!) support (can be toggled, as all parameters)
- Links can be numbered and accessed by number - as in lynx, but this
can be toggled during runtime
- Support for SSL and form history
- Javascript support (as a compile-time option) is on the way
- High configurability/customizability, including all keyboard
shortcuts, through dialogues within the browser
And the best thing is: Despite all these features, the elinks executable is still smaller than that of lynx (724k vs. 1.1 MB on my machine) and links (902k).-F
In Germany and Austria, prices for new books are fixed, i.e. a new book
always costs the same in every bookstore, including web
bookstores like Amazon, so that bookstores don't compete on price.
It does, however, not apply to (a) used books, (b) books
where the rest of an edition is being discounted by the
publishers, (c) books imported from other countries, (d) any other media, like music CDs, DVDs, software etc..
This law was made to protect small bookstores with chiefly literary,
cultural and academic programs from the competition of bookstore
chains (like Barnes & Noble in the U.S.) and mail order bookstores.
While one might have different opinions about free markets and free
pricing, this system indeed works as intended. Unlike in other Western
countries, Germany and Austria benefit from a wealth of small quality bookstores
in every town. In addition, there exists - since decades - a very efficient
national book wholesaler system,
so that any bookstore, regardless the size, can get any available book
for a customer usually on the next day (if it's not on stock in the
store already). Despite all this, Amazon still managed to
establish a hugely profitable business in Germany for various reasons -
the comfort of browsing an online catalogue and because
they offered, for the first time in Germany, an efficient way of ordering
English-language books.
The court decision simply maintains the fixed book price law for Ebay sales of
new books by commercial traders. It does not apply to Ebay sales of
second-hand books.
-F
However, I gave up on that project simply because, unlike for example in the case of gif vs. png images, there is no easy replacement to be advertised and offered to non-technical people:
In other words, a format that is open across applications and platforms, sufficiently powerful in its encoding both of typographic (font settings etc.) and structural (footnotes etc.) layout and widely supported by mainstream word processors (and be it only everything but MS Word) doesn't exist. As long as this doesn't change, for example with OASIS' current efforts to standardize an open office document format or large cross-application support for the OpenOffice file format, any "no-doc" advocacy is elitist and doomed to alienate even people who might be sympathetic for political reasons.
But if anyone wants to seriously do a grassroots campaign against using Microsoft's proprietary file formats, I am happy to transfer the no-doc.org domain to them for free.
Not if you use "enterprise" distributions, rely on support contracts, need staff training, inhouse development for migrating data and applications, or if you rely on proprietary software that runs on top of the OS (the way the film industry uses GNU/Linux as graphics workstation OS, for example).
But it's true that desktop GNU/Linux is good enough to have commoditized the product "desktop operating system" as it is defined by the standard system CD of MS Windows or MacOS X. Today, buying Windows licenses only makes sense if one needs something for running software on top of it that doesn't exist for GNU/Linux. Unlike in the days of the early MacOS or Windows 3.x, the raw desktop for itself has no proprietary product value anymore because KDE and Gnome have commoditized it. With the exception of Web browsers, GUI applications however are hardly commoditized yet. It may take a very long time until this point is reached, or it even might not be reached at all because of outside obstacles like software patents (on multimedia codecs etc.) and DMCA-imposed interoperability restraints.
- Distros like Xandros "just work"
Can be said about Windows 2000 and XP as well.
- Linux is secure from worms, trojans, viruses
This is a dangerous delusion. GNU/Linux has none of that security by design, but rather by the obscurity of not being a highly uniform mainstream desktop OS. There have been enough security holes in critical components of GNU/Linux (including the kernel itself) for malware to do as much harm as under Windows, and I guess a skilled attacker can crack almost any non-professionally administrated Linux box.
- Linux runs on modest hardware
Not if you run KDE, Gnome and Openoffice (as in a desktop system). In fact, a Windows 2000-based desktop system is faster and resource-friendlier in comparison.
- Linux is less complex and thus more stable
Nonsense. The combo of Linux kernel + OS daemons + GNU/shell userland + XFree86 + Gtk/Qt + KDE/Gnome middleware is just as complex as Windows.
- Linux has a "cool" factor missing from Windows
No argument.
- The IT world's view of Microsoft as "evil" is percolating down to the general public
No argument.
- Linux now comes with a sufficient set of applications for most common purposes
No, if you speak of the desktop. You can deploy a Linux-based desktop only in very well-defined (or limited) usage scenarios:
- Use of Internet clients (Mozilla): fully equivalent to Windows-based solutions
- Office suite (OpenOffice vs. MS Office): only against tradeoffs in Microsoft compatibility, speed, features and desktop integration
- Graphic design (Gimp vs. Photoshop): only against tradeoffs, suitable only for screen design
All other free desktop applications which currently exist for GNU/Linux are either small utilities (CD burning GUIs and the like), or they are ambitious, but not yet complete and not available in production quality (Scribus, Ardour etc.).
- Linux applications are more stable and simpler than Windows' ones
You don't mean the shell and the commandline userland, don't you? Which applications do you mean if you speak of the desktop, and not of mutt, vim, slrn etc.? (The GUI applications mentioned before are all cross-platform *nix/Windows btw.)
What you write are the clichés Windows people have in mind when they switch to GNU/Linux. Most of the rely on a false extrapolation from GNU/Linux as a Unix-like commandline and server operating system (which, if properly administrated, indeed is rock-solid and resource-efficient) onto GNU/Linux as a desktop OS.
-F
Also, KWord is built on top of KDE's component/toolkit architecture that is a world apart from MacOS X Carbon/Cocoa API. While Qt allows a native port to Aqua, it does not offer a native port to Carbon or Coca, and Apple is unlikely to establish a third desktop API on its platform just for the sake of getting a functionally rather limited word processor that, at the moment, has no dramatic advantage over the old Claris/AppleWorks offering.
And keep in mind that for Safari, Apple just used the engine (KHTML) of a free program, not the GUI application (konqueror) itself, in the same way it put its own (proprietary) GUI on top of Mach and BSD. From I experience, I doubt that KWord and Abiword are, in their present state, as attractive as "engines" as BSD and KHTML were. If it all, Abiword seems a more likely candidate since it's designed as a cross-plattform application and, quite in opposition to KWord, focuses on getting base functions and usability right before acquiring more nifty/hackerish features such as frame-based page layout and importing PDF files.
What makes your scenario very unlikely in the end are licensing issues. KOffice and Abiword are GPLed code and thus would require Apple to release any program based on them under the GPL. Which doesn't fit to the company's successful tactics of putting slick, but proprietary GUIs on top BSD- or LPGL-licensed hacker code like BSD and KHTML. A GPLed "iWord" that could be ported back to Linux and even Windows would, unlike the current i-apps, be no exclusive selling point for MacOS X.
-F
Well, KWord has a WordPerfect import filter.
> I'm happy with my Linux system right now. It supports all my hardware
> and gives me a nice desktop.
Then there's no reason for you to switch. If you would instead not be
happy with your system and find that FreeBSD runs better on your
hardware, than this would be a reason to go through the hassle of
switching your OS.
After all, the differences between a GNU/Linux and a *BSD system are
practically user-invisible on the level of the desktop interface. Both
are, as a matter of fact, state-of-the-art Unix, but may not be called
so because of trademarks and expensive certifications (which, contrary
to popular make-belief, are not owned by SCO, but by the Open Group,
formerly X/Open).
The differences mainly concern the kernel (partitioning schemes and
filesystems, hardware drivers, module handling, packet filters, sound
and multimedia subsystems). Userspace differences in the init system,
package management and base OS/distribution tools are not bigger than
those between two GNU/Linux distributions. Slackware or Gentoo users
might even find Free/Net/OpenBSD more familiar than RedHat or SuSE.
There are subtle, but sometimes crucial differences in the commandline
userland between GNU/Linux and *BSD though, unless one installs the GNU
file and text utilities on *BSD and uses them as default (which is easy
and supported by the package management of all three free BSDs). The
KDE/Gnome/XFCE desktops act in a completely transparent manner, with no
visible differences, on top of GNU/Linux and *BSD.
Generally, the Linux kernel is best suited for a desktop system because
of its more advanced sound (ALSA) and video (video4linux) subsystems,
support for a wide range of desktop hardware gadgets (Webcams, graphic
tablets and the like), filesystems (including everything from Amiga to
Acorn...) and hardware-accelerated video card drivers (DRI/DRM, although
I read similar features are on the roadmap of FreeBSD and NetBSD), and,
since kernel 2.6, kernel preemption and low-latency functions. However,
FreeBSD should come close to Linux as it is optimized for the x86-PC
architecture and is a very good performer. NetBSD is, from my own
experience of running it as a secondary OS, not as fast, but still
surprisingly good for an OS that is developed with portability (and
hence abstraction/clean interfaces vs. optimization) as its prime
objective. IMHO, it is (very) roughly comparable with Linux kernel v2.0
in terms of performance and desktop computing friendliness. Installation
of NetBSD is a bit difficult (more so than even Debian), and the
necessity of creating classical BSD disklabels for every storage media
to be mounted can be highly annoying on a desktop system (for example,
if one wants to quickly mount someone else's USB stick). OpenBSD is,
IMHO, a bad choice for a desktop system unless security and crypto
features are the main requirements. It is not a good performer at all
(and not being developed with performance as a main goal).
Kernel-wise, FreeBSD's chief advantage over Linux used to be better
responsiveness under high system load and better virtual memory
management (which both gave/give FreeBSD an edge over Linux on servers
rather than on desktops). This advantage has gradually decreased through
substantial low-level improvements in Linux 2.0, 2.4 and now 2.6, which
AFAIK has lifted quite a bit from FreeBSD's advanced VM management.
Maybe Linux 2.6 is now on par, but still I wouldn't be surprised if
FreeBSD (and also NetBSD) would be more mature in this field. (For
example, I never succeeded in bringing down my two NetBSD boxes with a
fork bomb.)
-F
It would make more sense IMHO to abolish file selectors altogether and instead throw users into their preferred file manager for opening files. All it would need is a freedesktop.org standard protocol for file manager/application interaction and perhaps a $FILEMANAGER environment variable. (Theoretically, $FILEMANAGER could then also be a shell in a terminal.)
-F
For those who haven't heard of it yet: TEI is an open SGML/XML format created for electronic editions of literary texts. It is as comprehensive and well-designed for text philology as DocBook is for technical documentation. The only drawback is that it is, like DocBook, very comprehensive and accurate in its markup tags (fulfilling all needs of academic editions of historical texts), so that for average readers, the trimmed-down TEI Lite DTD should do the job.
For e-literature collections created by professional philologists - such as the Victorian Women Writers Project, TEI already is the standard text format. Thanks to the SGML/XML toolchain, TEI sourcecode can, like DocBook, of course be painlessly transformed into HTML, txt, RTF, PDF etc. (TEI is, btw., also being mentioned in Eric S. Raymond's quite useful DocBook Demystification HOWTO.)
Florian
(philologist by profession)
If UserLinux was an end user-oriented distribution, it surely had to pick KDE instead of Gnome, since KDE is the more integrated and stable GUI and is less messy in the architecture underneath (while Gnome/GTK has the lead in 3rd-party applications and, since recently, UI polish).
But for a "Free Enterprise Linux", there must not be any hidden costs for enterprise software development. This demands that libraries and SDKs should, where possible, be LGPL- or BSD-licensed, and not GPLed with for-pay-exceptions (like in Qt and MySQL).
Of course, the question remains if, due to its proprietary-friendly licensing and relatively conservative (=stable) design process, FreeBSD wouldn't be the better "Enterprise Linux" anyway. After all, the GPLed Linux kernel could be ditched in favor of a BSD kernel with almost the same arguments the UserLinux project now ditched the GPLed KDE libraries in favor of the LPGLed Gnome libraries.
But since Linux is all the hype even where it doesn't make too much sense (like in PDAs, for which Minix would be much better suited), it's good that the "UserLinux" project attempts to prevent that commercial distributors do the same horrible mistakes with Linux and their "enterprise" distributions the proprietary Unix vendors made in the 1990s.
-F
The title "Linux Power Tools" falsely suggests that the book is a sequel or update to O'Reilly's "Unix Power Tools". Unlike the book reviewed here, "Unix Power Tools" is not about configuring KDE and such stuff, but a wizard's guide to & treasury of classical shell tools, arguably the best Unix user book ever written, the bible of the commandline, the ultimate celebration of design philosophy genius behind Unix. In that light, calling the reviewed book "Linux Power Tools" is pure blasphemy.
-F
-F
Fortunately, I was able to exchange the BTC against a marginally more expensive LG drive (4040B). While it burns CDRs only at 24x and CDRWs at 12x, it worked out of the box and like a charm with all programs under GNU/Linux.
A nice plus of the LG is that, besides DVD+-R[W] and CDR[W], it also reads and writes DVD-RAM. Unlike the other writable DVD and CD formats, DVD-RAM is true random access optical removable storage (similar to an MO cartridge, not just a userspace simulation of random access like Mount Rainier packet writing). It works as standard IDE device without any specific drivers or software frontends, so that DVD-RAM media can be partitioned and formatted with ext2/xfs/reiserfs under Linux just like a removable hard drive. Since it is possible, for example, to rsync directories to DVD-RAM and since the media are affordable (~$13 for a double sided 2x4,7 GB disk), it strikes me as a near-ideal backup solution.
-F
However, it was added without patching the wm itself and thereby bloating its code. Instead, since ratpoison can be fully controlled via the commandline, the "expose" functionality (to be found in the "contrib" directory as "rpshowall.sh") was written as an external shell script which tells ratpoison to split frames in a certain way. Through ratpoison's freely definable keybindings, the script can be used like a built-in function/command of the wm.
Conventional, but probably true answer: servers. There are still many companies running standard services like mail, web etc. on proprietary operating systems (Sun, Microsoft) in a time where it makes no whatsoever sense anymore. With kernel 2.6, Linux will gain acceptance as a high-end Unix replacement and be deployed wherever older server installations need to be replaced.
No. The desktop UI is still too inconsistent across KDE/Qt, Gnome/GTK, Mozilla/XUL and Openoffice and still offers no viable alternative to the commandline when it comes to system administration/configuration.
I predict that in 2004, attention will move away from KDE and Gnome as all-in-one-solutions. Instead, it will be finally accepted as reality among developers and users that different GUI APIs will continue to coexist, and that efforts should be made to standardize the protocols and user interfaces across the APIs. For the future of GNU/Linux and *BSD on the desktop, freedesktop.org will be much more important than kde.org and gnome.org, but it could take five-ten years until the difference between a KDE/Qt, GTK/Gnome, Tcl/Tk, Fltk program will be as irrelevant to users as the difference between a Carbon and a Cocoa app on MacOS X or as that between a Microsoft MFC program written in C++ or an OWL-based program written in Borland Delphi for a Windows user.
Once this level of standardization is reached, the importance of all-in-one desktops like KDE and Gnome could dramatically decrease, since users instead could combine components like taskbars, window managers, file managers and system menus at will. (Which, thanks to freedesktop.org, is already possible: fspanel / fbpanel / suxpanel / the xfce4 panel can be used as drop-in replacements for the Gnome panel, rox / xffm4 as drop-in replacements for Nautlius, and the list of freedesktop.org-compliant window managers suitable as replacements of metacity / kwin is endless.)
However, it will take yet another five years until 2013 or 2014 that a standardized Unix/GNU/Linux/BSD desktop will allow developers of system components (like sendmail/exim/Postfix/qmail, lpr/cups, Samba, Grub/Lilo...) to write GUI configuration panels for their own software. At the moment, desktop projects like KDE, Gnome/Ximian and Webmin can only provide insufficient configuration wrappers around low-level system tool; the only sane solution is that such GUI configuration panels are provided by the original component developers in sync with their release schedules, and will work consistently on any GUI configuration (as opposed to the present situation where a configuration panel would have to be provided in separate versions for KDE, Gnome, XFCE, webmin and what-have-you). Only at this point, GNU/Linux will be able to replace commercial end-user GUI operating systems on a large scale and be accessible to home users.
Contrary to what Eric S. Raymond says: The unclear situations of RedHat/Fedora and SuSE (after it has been bought by Novell) could create a strong push towards Debian as the standard binary (GNU/)Linux distribution. The Debian core distribution could become a de-facto-replacement of the disappointing "Linux Standards Base (LSB)", as more and more (commercial and community) distributions will be based Debian. Knoppix, Lindows and, in the near future, User Linux are prime examples. Debian itself will gain more acceptance in the mainstream and among new users as soon as it will ship with the new installer.
Given the record of Netscape/Mozilla's, StarOffice/OpenOffice's and Apple Darwin's transformation of corporate into public development projects, I doubt that RedHat/Fedora will ever become a true community project. It is also being overlooked that the equation RedHat=Linux is specific to the U.S. only
- The Mach+BSD server design is a kludge creating unneccessary bloat,
complexity and performance overhead without exploiting any of the
potential advantages of a microkernel design like better portability or
Hurd-style hack value like filesystems running as daemons in userspace
etc.
- In any case, Linux 2.4 and all the more 2.6 should beat, in terms of
performance and scalability, the crap out
of MacOS X' combination of vintage Mach with vintage BSD and a bloated
GUI on top
- Debian and NetBSD don't have compatibility bloat like the "Classic"
virtual machine, m68k-CPU-Emulation and "Carbon"-API in MacOS X
- They have much cleaner filesystem layouts
than OS X with its inconsistency of Unix directories (/bin,
/etc) which
are hidden on the GUI level and application folders inherited from
NextStep
- They have a more consistent and robust configuration system than
MacOS X with its horrible Registry-like "Netinfo" database that
replaces some, but not all configuration files in
/etc
- They come with a more complete (and especially in Debian's case
thanks to GNU) powerful set of classical Unix commandline applications
- For the software which is not installed by default, they have consistent package management while MacOS X has a
number of simultaneous/incompatible package managers and databases which
don't know each other's dependencies: MacOS X install images, fink,
GNU/Darwin, BSD-style pkgs/ports...
- They install programs like vim, mutt, shells etc. with sensible default
configurations while I find the commandline userland and MacOS X almost
unusable they way it is configured out of the box
For someone who primarily works on the commandline and needs graphical programs only for the occasional web browsing, graphics/pdf and video viewing (for all of which excellent free, X11-based solutions like mozilla-firebird and mplayer do exist), MacOS X offers no advantages over a GNU/Linux or NetBSD system in which all the system- and commandline-level things are done cleaner and better. So it seems Bill Joy doesn't write in vi and work the Unix way anymore, otherwise he would have better things to say about Linux.This comment must have been written by someone (and modded up by
someone else) who doesn't know much about XFCE. XFCE is not just a
window manager, but a fully integrated, mouse-configurable desktop
including wm, panel (with panel applets), taskbar, pager and graphical
file manager (including Samba browsing support), extensive drag'n'drop
capabilities including for printing, central configuration menus
including sound and mouse setup. It's based on Gtk 2.x and
freedesktop.org standards (and thus with a high degree especially of
Gnome and KDE interoperability).
Think of XFCE as a desktop environment without the redundant middleware
layers (DCOP/KParts/arts etc. in KDE, Corba/Bonobo/esd/Gconf etc.i in
Gnome) that make both KDE and Gnome bloated and slow, and which are
hardly used by third party applications outside main Gnome and KDE
distributions at all.
So it amazes me that the previous commentator thinks XFCE is "not a
desktop". On the contrary, XFCE is a desktop done architecturally right,
similar to, for example, the desktops of AmigaOS, RiscOS, Macintosh
Classic and BeOS. While both the XFCE panel (with its legacy to the user
interface of the CDE panel) and new file manager could still need some
usability improvement, the architectural foundation is excellent.
XFCE is also the proof that a X11- and GNU/Linux-/BSD-based desktop
computer can be as fast and efficient as one would normally expect from
a Unix-like system. In other words, it's as fast as a basic window
manager setup with Window Maker/icewm/fvwm2 while providing a fully
integrated desktop that doesn't require users to run the shell or edit
configuration files.
(A prominent XFCE user and supporter is, btw., Alan Cox.)
(Offtopic: A similarly nice, elegant solution for desktop/clients PC printing is pdq, which unlike lpd and cups runs only as a local spooler without opening a network port, and is lean (65k), dead-simple and functional. With nullmailer/ssmtp & pdq, I managed to close all ports (except of course SSH) on my two desktop PCs under Debian GNU/Linux without any firewalling. AFAIK, Debian is the only OS offering all the aforementioned pieces of software as part of its main distribution.)
Translation (German original contains a number of colloquial expressions which don't translate 1:1 into English): "What can I say? A top-notch piece of engineering. I am drooling!"
> Similarly, if you like the OPL, keep using it. It's still a perfectly valid, legal license.
Yes, but unmaintained legal code is as problematic as unmaintained program code. No organization will enforce these licenses or, if necessary, defend them in court; nobody will update them if new legal or technical conditions make it necessary (as in GPL v2.0 vs. GPL v1.0 vs. the upcoming GPL v3.0). Which renders opencontent.org's licenses worthless IMHO.
It is especially not funny to see the Open Publication License go away. It had a considerable momentum among book publishers - being used, among others, by O'Reilly and the Bruce Perens book series of Prentice Hall. I myself put all my papers under the OPL, encouraged other people to do so as well, and now feel severly f*cked and betrayed by this move. The instability and unreliability now associated with open content copylefts could severely damage the whole movement. As someone who managed to convince a large German public library to release its online content under the Open Content License, I am severely pissed & awaiting to take the beating for opencontent.org's irresponsibility.
The Creative Commons licenses, in my view, are not an alternative because they are too many and incompatible to each other, thus creating confusion and preventing exchange between work copylefted under its terms. What's still worse is that most Creative Commons licenses are not free in the sense of the Free Software definition of the FSF, the Debian Free Software Guidelines or the Open Source Definition.
I urge the initiator of opencontent.org to keep the website alive, and if only as a central link repository to other sites, and provide a smooth/sensible upgrade path from the Open Content License and the Open Publication License to particular Creative Common Licenses, for example by developing a license which would simultaneously be "Open Publication License v2.0" and "Creative Commons License foo". Given the amount of work that already circulates under either the Open Content License or the Open Publication License, anything else would be utterly irresponsible.
Imagine the FSF suddenly abandoning/stalling the GPL in favor for someyet-unwritten different license, leaving ten thousands of Free Software developers in the legal lurch & betraying their trust. What is an unlikely horror scenario for free software is now the reality of open content.
Bravo, opencontent.org, Microsoft, the RIAA, the MPA, SCO and all other old copyright regimes now have another reason to cheer and point at copyleft culture as immature, unreliable, not viable for serious publishing, etc.. Please wake up and release that you have taken up a responsibility which you cannot so easily throw away!
Do new consumer cards solve this problem, or do they layer up useless extra features?
A major improvement in FreeBSD 5.x over 4.x is the new modular init. Instead of one monolithic script (classical BSD) or several scripts in a symlink farm with manual sorting and dependency resolution (SysV / Debian, RedHat, SuSE...), it uses an internal automatic sorting and dependency resolution comparable to apt-get or modprobe on GNU/Linux. I would like to see mainstream adoption of this in the GNU/Linux world of this. To date, Gentoo Linux is the only distribution offering and supporting this excellent feature.
Since Minix has been put under the BSD license since April 2000, I wonder why nobody has made an effort yet to port it to embedded systems (PalmOS PDAs with Dragonball CPUs, for example, should be an ideal target). Minix should be much better suited for many embedded applications than the much more complex Linux kernel.